Bugfix

french_cities/city_finder.py

@@ -250,10 +250,12 @@ def find_city(
     # cities (using dep & city) using fuzzy matching
     ix = addresses[addresses["candidat_0"].isnull()].index
     if len(ix) > 0:
-        missing = addresses.loc[ix, [dep, "city_cleaned"]]
+        missing = addresses.loc[ix, [dep, "city_cleaned"]].rename(
+            {dep: "#dep#"}, axis=1
+        )
         missing = _find_from_fuzzymatch_cities_names(
             year, missing, "candidat_missing"
-        )
+        ).rename({"#dep#": dep}, axis=1)
         addresses = addresses.merge(
             missing, on=[dep, "city_cleaned"], how="left"
         )
@@ -319,6 +321,7 @@ def list_map(df, columns):
             session=session,
             rename_candidat=f"candidat_{k+1}",
             addresses=addresses,
+            dep=dep,
         )
 
         if type_ban_search == "municipality":
@@ -348,6 +351,7 @@ def list_map(df, columns):
                     session=session,
                     rename_candidat=f"candidat_{k+1}",
                     addresses=addresses,
+                    dep=dep,
                 )
 
     # Proceed in two steps to keep best result (in case there are results from
@@ -555,8 +559,8 @@ def _find_from_fuzzymatch_cities_names(
     df = df.reset_index(drop=False)
 
     results = []
-    for dep in look_for["dep"].unique():
-        ix1 = look_for[look_for.dep == dep].index
+    for dep in look_for["#dep#"].unique():
+        ix1 = look_for[look_for["#dep#"] == dep].index
         ix2 = df[df.dep == dep].index
         match_ = pd.DataFrame(
             cdist(
@@ -567,15 +571,14 @@ def _find_from_fuzzymatch_cities_names(
             ),
             index=ix1,
             columns=df.loc[ix2, "CODE"],
-            # index=look_for.loc[ix1, "city_cleaned"],
-            # columns=df.loc[ix2, "TITLE_SHORT"],
         ).replace(0, np.nan)
-        # print(match_)
 
         try:
             results.append(
                 pd.Series(
-                    match_.dropna(how="all", axis=1).idxmax(axis=1),
+                    match_.dropna(how="all", axis=1)
+                    .dropna(how="all")
+                    .idxmax(axis=1),
                     index=ix1,
                 )
             )
@@ -867,6 +870,7 @@ def _filter_BAN_results(
     session: Session,
     rename_candidat: str,
     addresses: pd.DataFrame,
+    dep: str = "dep",
     fuzzymatch_threshold: int = 80,
     ban_score_threshold_city_known: float = 0.6,
     ban_score_threshold_city_unknown: float = 0.4,
@@ -890,6 +894,9 @@ def _filter_BAN_results(
         Columnn to rename the results to.
     addresses : pd.DataFrame
         Full DataFrame of address to store the kept results into.
+    dep : str, optional
+        Field (column) containing the department values. Set to False if
+        not available. The default is "dep".
     fuzzymatch_threshold : int, optional
         The fuzzy match score threshold (on city labels) to keep the results.
         Default is 80.
@@ -910,7 +917,7 @@ def _filter_BAN_results(
     results_api = find_departements(
         results_api, "result_citycode", "result_dep", "insee", session
     )
-    ix = results_api[results_api.dep == results_api.result_dep].index
+    ix = results_api[results_api[dep] == results_api.result_dep].index
     results_api = results_api.loc[ix]
 
     if results_api.empty:

french_cities/departement_finder.py

@@ -342,7 +342,7 @@ def find_departements_from_names(
             process.extractOne(
                 x,
                 candidates_keys,
-                scorer=fuzz.token_set_ratio,
+                scorer=fuzz.ratio,
                 score_cutoff=80,
             )[0]
         ]

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "french-cities"
-version = "0.3.2"
+version = "0.3.3"
 description = "Toolbox on french cities: set vintage, find departments, find cities..."
 authors = ["thomas.grandjean <[email protected]>"]
 license = "GPL-3.0-or-later"

tests/test_departement_finder.py

@@ -20,8 +20,15 @@
 
 input_df2 = pd.DataFrame(
     {
-        "deps": ["Corse sud", "Alpe de Haute-Provence", "Aisne", "Ain"],
-        "codes": ["2A", "04", "02", "01"],
+        "deps": [
+            "Charente-Maritime",
+            "Seine-et-Marne",
+            "Corse sud",
+            "Alpe de Haute-Provence",
+            "Aisne",
+            "Ain",
+        ],
+        "codes": ["17", "77", "2A", "04", "02", "01"],
     }
 )